Project Summary Protein-protein interactions are involved in nearly every cellular process yet defining which proteins interact with one another has been challenging. Many of these interactions are dictated by domain that interaction with short linear amino acid sequences. These domains have been conserved across Archaea, Bacteria, and Eukaryota. In Human there are over 1000 proteins that use one of these domains to interaction with other proteins. While many of these domains have been studied we have failed to produce a predictive code of their peptide specificity that would include the functional consequence of mutations. This inability to provide a predictive model is true for one of the most common of these domains in human, the PDZ domain, and many mutations within these domains and their targets have been associate with a variety of diseases. In addition, the PDZs of the human microbiome have been largely ignored because of the misconception that these domains are more prevalent in Eukaryotes. While this is true on an organism by organism basis, there are actually more total PDZ domains in the 100 most common microbes of the human microbiome than all of the human PDZs combined. As disruption of the microbiome has been associated with multiple diseases, these domains and the pathways they control may provide critical insight to the health of the microbiome and the human host. The goal of this work is to provide a predictive understanding of the PDZ domain and its target preference. Long-term we hope to establish this approach as a blueprint method leading to models for all peptide-interacting domains and provide immediate understanding of the consequence of a mutation found in the domain or its targets. Using a newly developed hybrid assay that is sensitive, simple, and high throughput we will first characterize the target preferences of all human PDZ domains. This method captures a greater dynamic range than prior methods and in preliminary work produced more predictive data than prior approaches. Our second Aim is to then characterize all of the PDZ domains of the human microbiome as these represent more divergent domains and have the potential to have a large impact on human health. Finally, we will investigate variation found in human domains associated with disease as well as take a synthetic approach to engineer and understand the domain?s rules of peptide recognition. Together we hope to comprehensively explore the domain and its binding capacity. As genome sequencing becomes a common medical diagnostic, our goal is for our model to be used by the community to understand the potential consequences of any mutations found in the coding sequences of these domains.